Jeff Bezos Backs Flourish, a $500M Brain-Inspired AI Startup Chasing a 50-Watt Path Past AI's Energy Wall

A New York startup called Flourish has raised $500 million in its first outside round at a $2.5 billion valuation, with Jeff Bezos personally anchoring the deal — and a thesis that runs directly against the grain of the current AI boom: that the way past the industry's energy wall is to reverse-engineer the human brain, not to keep stacking silicon. The round, which closed around June 4, drew in Lux Capital, GV (Alphabet's venture arm), and Catalio Capital alongside Bezos.

Bezos's involvement grew as the round came together. He initially committed roughly $50 million, then nearly doubled his stake to about $100 million after other high-profile backers piled in — a vote of confidence that helped set the company's headline valuation before it has shipped a product. It is a separate bet from his physical-AI lab, Project Prometheus, underscoring how aggressively the Amazon founder is now spreading capital across frontier approaches to intelligence.

The founding team is unusually pedigreed. Flourish was co-founded by Thomas Reardon and Rob Williams. Reardon created Internet Explorer at Microsoft in the 1990s, then founded the brain-computer interface company CTRL-labs, which Meta acquired in 2019 for an estimated $1 billion; he went on to lead neuromotor-interface work at Meta Reality Labs, including the Neural Band wristband. Williams is a former Amazon "S-team" executive, part of the company's most senior leadership tier.

The company's system, Cortex AI, aims to emulate brain function by mapping real neurons and their connections — the field known as connectomics — to find what Flourish calls the brain's "core algorithm," rather than scaling conventional neural networks ever larger. The pitch is built around efficiency: Flourish says Cortex AI is designed to run at 20 to 50 watts, roughly a laptop's power draw, instead of the megawatt-scale data centers that today's frontier models demand.

That framing lands at a pointed moment. Even as labs sign multibillion-dollar compute contracts and the largest IPO in history queues up around AI infrastructure, power availability has become the binding constraint on scaling. Flourish is wagering that biology already solved the efficiency problem — the human brain runs on about 20 watts — and that decoding how it does so is a faster route to capable, affordable AI than another generation of accelerators.

The risk is equally clear: neuroscience-inspired computing has a long history of bold promises and slow payoffs, and connectomics remains an enormous scientific undertaking. Flourish now has half a billion dollars, a marquee cap table, and a founder who has twice built things the rest of the industry adopted — but no shipping model yet to prove the brain's blueprint can be turned into working software.